Sex in Penicillium: Combined Phylogenetic and Experimental Approaches

Fungal Genetics and Biology 47 (2010) 693–706

Contents lists available at ScienceDirect

Fungal Genetics and Biology

journal homepage: www.elsevier.com/locate/yfgbi

Sex in Penicillium: Combined phylogenetic and experimental approaches

M. López-Villavicencio a,*,1, G. Aguileta b,1, T. Giraud b, D.M. de Vienne a,b, S. Lacoste a, A. Couloux c, J. Dupont a a Origine, Structure, Evolution de la Diversité, UMR 7205 CNRS-MNHN, Muséum national d’histoire naturelle, CP39, 57 rue Cuvier, 75231 Paris Cedex 05, France b Ecologie, Systématique et Evolution, UMR 8079, Bâtiment 360, Université Paris-Sud, F-91405 Orsay cedex, France; UMR 8079, Bâtiment 360, CNRS, F-91405 Orsay cedex; France c Genoscope – Centre National de Séquençage: BP 191, 91006 EVRY cedex, France article info abstract

Article history: We studied the mode of reproduction and its evolution in the fungal subgenus Penicillium Biverticillium Received 25 January 2010 using phylogenetic and experimental approaches. We sequenced mating type (MAT) genes and nuclear Accepted 6 May 2010 DNA fragments in sexual and putatively asexual species. Examination of the concordance between indi- Available online 9 May 2010 vidual trees supported the recognition of the morphological species. MAT genes were detected in two putatively asexual species and were found to evolve mostly under purifying selection, although high sub- Keywords: stitution rates were detected at some sites in some clades. The ﬁrst steps of sexual reproduction could be Positive selection induced under controlled conditions in one of the two species, although no mature cleistothecia were Relaxed selection produced. Altogether, these ﬁndings suggest that the asexual Penicillium species may have lost sex only dN dS very recently and/or that the MAT genes are involved in other functions. An ancestral state reconstruction Talaromyces analysis indicated several events of putative sex loss in the genus. Alternatively, it is possible that the Experimental crosses supposedly asexual Penicillium species may have retained a cryptic sexual stage. Ó 2010 Published by Elsevier Inc.

1. Introduction appear to have had multiple transitions from sexuality to asexuality (Lobuglio et al., 1993). In Fungi, asexual reproduction by pro- Despite the costs of sex (Otto and Lenormand, 2002), most duction of asexual propagules (e.g. conidia) has been considered eukaryotes engage in sexual recombination at least at some point to be particularly common, with a quarter of fungal species in their life cycle. The predominance of sexual reproduction sug- thought to reproduce only by asexual means (Taylor et al., 1999). gests that sex must provide some advantages. Although asexual However, recent studies have shown that a great number of fungal reproduction is common in nature, exclusively asexual taxa are ‘‘asexual” species are in fact capable of sexual reproduction. Sex in rare and they are considered to be short-lived (Judson and these species is only difficult to observe in nature and challenging Normark, 1996). Many models have been built to explore short- to induce in the laboratory. Some species for which sex has not and long-term advantages of sex to explain its maintenance, but been observed, like Coccidioides immitis, present population struc- evidence from natural cases is still scarce (but see De Visser and tures consistent with recombination, suggesting the existence of Elena, 2007, for recent experimental evidence for direct benefits cryptic sex in nature (Burt et al., 1996). Sex has been successfully of sex). More information is needed on suitable biological models induced under controlled conditions in species such as Candida that could be used to tackle these issues. albicans (Hull et al., 2000), Aspergillus fumigatus (O’Gorman et al., The empirical study of sex and recombination has been based 2009), A. flavus and A. parasiticus, which were long thought to be on vertebrates, insects and plants models, while other groups of asexual (Horn et al., 2009a,b). Finally, apparently functional mating eukaryotes, including fungi, have been neglected (Birky, 1999). type genes (i.e. genes that define mating compatibility in fungi) Biological groups that exhibit a diversity of reproductive strategies have recently been identified and characterized in several species provide unique opportunities to study the evolution of sex. Groups with no sexual cycle described, such as A. oryzae (Galagan et al., such as fungi are thus excellent models as they present a great 2005), and recently in Penicillium chrysogenum and Acremonium range of reproductive strategies, including obligatory sexual spe- chrysogenum (Hoff et al., 2008; Pöggeler et al., 2008). Other fungal cies, those that alternate sexual and asexual reproduction, and oth- species shown to be truly clonal from a population genetic stand- ers that appear to be strictly asexual (Taylor et al., 1999). Fungi also point, such as Penicillium marneffei, also present mating type genes, suggesting that sex has been lost recently (Woo et al., 2006; Fisher, 2007). Alternatively, sexual reproduction that would always occur between identical clones, as allowed under homothallism, would * Corresponding author. Fax: +33 1 69 15 73 53. E-mail address: [email protected] (M. López-Villavicencio). not be distinguishable from strictly asexual reproduction using 1 Both authors contributed equally to this paper. population genetics.

1087-1845/$ - see front matter Ó 2010 Published by Elsevier Inc. doi:10.1016/j.fgb.2010.05.002 694 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

Besides the fact that Fungi represent suitable biological models mon descent. Most ascomycetes present two idiomorphs MAT 1- for studying the maintenance of sex and recombination, the study 1 and MAT 1-2. These genes code for transcription factors that in- of reproductive strategies in this group has important direct appli- duce the production of pheromones and pheromone receptors. The cations. Until very recently, many fungal plant pathogens and MAT 1-1 idiomorph includes a gene encoding a protein with a mo- some important animal pathogens and species with biotechnolog- tif called the a1 domain, while the MAT 1-2 idiomorph presents a ical importance were assumed to be clonal with widespread distri- gene encoding a protein with a DNA-binding domain similar to butions (Taylor et al., 1999). However, recent studies have revealed that of the high mobility group (HMG) (Coppin et al., 1997). Homo- the existence of recombination in some of these important species thallic fungi can undergo intra-haploid mating (Giraud et al., (Burt et al., 1996; Couch et al., 2005). Deciphering the mode of 2008), the proximal cause being in most of the filamentous homo- reproduction of pathogens may have profound implications for thallic ascomycetes, each haploid possesses two alternate forms of our understanding of the biology and for the management of the the MAT locus in its genome (Coppin et al., 1997). In contrast, het- species. Recombination should maintain genetic variation within erothallic fungi carry a single MAT idiomorph and two strains car- populations and generate new genotypes, which can present new rying complementary MAT idiomorphs are required for sex to or increased virulence, pathogenicity or drug resistance (Dyer occur. and Paoletti, 2005; Taylor et al., 1999). Furthermore, many of the Selective pressures acting on mating type genes are expected to fungi used in industry are thought to have only asexual reproduc- be different in sexual vs. asexual species, and these pressures can tion, and the discovery of a sexual stage could help improving be detected based on sequences (i.e. O’Donnell et al., 2004). In sex- strain quality by crosses (Pöggeler, 2001). ually reproducing species, mating type genes must remain func- Here, we studied the evolution of reproduction in the fungal tional in order for sex to take place, implying that purifying subgenus Penicillium Biverticillium using phylogenetic and experi- selection should act (Devier et al., 2009; O’Donnell et al., 2004). mental approaches. This group includes important species, such On the other hand, in asexual species still carrying mating type as the opportunistic pathogen P. marneffei, food and feed spoilers genes, mutations driving loss of function should be selectively neu- as well as species of importance in the food and biotechnology tral (Fisher, 2007), leading to relaxed selection. Positive selection industries, such as P. pinophilum and P. funiculosum (Domsch et al., may act on MAT genes in sexual species, involving rapid and recur- 1980). In the fungal genus Penicillium counting ca. 250 species (Pitt, rent changes that proved advantageous (Wik et al., 2008). We 1979), only few species have a complete life cycle described, the therefore looked for footprints of acceleration in substitution rates teleomorphs being then Talaromyces or Eupenicillium. The remain- (indicating relaxed selection or positive selection) on MAT genes. ing species are considered as strictly asexual fungi, corresponding Finally, when species were found to have both MAT 1-1 and MAT to several independent losses of sex (Lobuglio et al., 1993). Asexual 1-2 alleles, we attempted to induce a sexual cycle experimentally. species have been classified into four subgenera, Aspergilloides, Biverticillium, Penicillium and Furcatum on the base of the morphol- 2. Material and methods ogy of their penicilli (Pitt, 1979). The subgenus Biverticillium is phylogenetically related to Talaromyces (Lobuglio et al., 1993), 2.1. Group of study and fungal isolates while the remaining three subgenera are related to Eupenicillium (Peterson, 2000) and are close to Aspergillus and related teleo- Most of the isolates of Talaromyces and Penicillium used in this morphs (Berbee et al., 1995). The teleomorphs Talaromyces and study were obtained from the LCP culture collection (Laboratoire Eupenicillium constitute two distinct phylogenetic lineages within Cryptogamie Paris) at the French Natural History Museum (Musé- the family Trichocomaceae. um National d’Histoire Naturelle), Paris. Additional isolates used We first constructed a robust phylogeny of the fungal subgenus were donated by the MUCL collection (Mycothèque de l’Université Penicillium Biverticillium to (1) improve our knowledge of their Catholique de Louvain, Louvain La Neuve, Belgium) and by CMPG relationships, (2) test if the criterion of concordance between mul- (Collection Mycologie Pharmacie Grenoble) and are now included tiple gene genealogies supported the extant described morpholog- in the LCP collection (Supplementary material Table S1). ical species (Dettman et al., 2003; Le Gac et al., 2007), and (3) evaluate the relationships between sexual and asexual species, in order to estimate the number of transitions towards asexuality. 2.2. Phylogenetic marker selection and primer design We used more strains per species, more genes, and partly different species, than those analyzed previously by Lobuglio et al. (1993). We used the FUNYBASE (Marthey et al., 2008) in order to search For building phylogenies, we used several genes recently proposed for single-copy orthologs and estimate their phylogenetic perfor- by Aguileta et al. (2008) as having a high phylogenetic performance mance at different taxonomic scales within the Penicillium genus. in fungi, in addition to some genes commonly used for fungal phy- We chose genes having an ortholog, and no paralog, in all complete logenies. Most of the published fungal phylogenies are indeed fungal genomes analyzed by Aguileta et al. (2008), having a high based on the same DNA markers, for instance the ribosomal genes phylogenetic performance (i.e. topological score higher than 91, or spacers, and genes coding for elongation factor proteins, RNA see Aguileta et al., 2008), and having different levels of divergence polymerase and beta tubulin (James et al., 2006; Lobuglio et al., among species. The exact functions of these genes are uncertain 1993). However, a recent study has shown that these genes may and their putative functions were inferred by BLAST analyses. We be suboptimal phylogenetic markers, inferring different relation- therefore only used their abbreviated names as in Aguileta et al. ships among species than those supported by most genes in the (2008) (for further details on the genes and their annotations see genomes (Aguileta et al., 2008). We therefore wanted to compare Marthey et al., 2008). Three genes were chosen: MS277 (putative the utility of the genes typically used in fungal phylogenies vs. that ribosome biogenesis protein), MS456 (putative DNA replication of the genes found to have a high phylogenetic performance by licensing factor), and FG610 (putative chaperonin complex compo- Aguileta et al. (2008). nent TCP-1). MS277 and MS456 were the two sole genes found to We were also interested in assessing whether the species con- yield the same topology as whole-genome sequences by Aguileta sidered as asexual were in fact so. For this goal, we first tried to de- et al. (2008). Using FUNYBASE, the protein sequences of the chosen tect mating type genes, and sequenced them in sexual and asexual genes were downloaded from A. fumigatus, as this species was one species. Mating-type loci are called ‘‘idiomorphs” rather than al- of the closest to the group Penicillium/Talaromyces available in the leles in ascomycetes due to the uncertainty of the origin by com- database and could therefore be used as an outgroup. To obtain the M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 695 nucleotide sequences, we performed TBLASTN searches at NCBI. 2.4. DNA extraction, amplification and sequencing of nuclear genes The recovered nucleotide sequence of A. fumigatus was also blasted (BLASTN) against the genome sequences of P. marneffei and Talar- Isolates were grown on MEA (2%) plates for 3–5 days at 25 °C. omyces stipitatus available in GenBank. The nucleotide sequences DNA extraction was performed using the DNeasy Plant Mini Kit from each candidate ortholog extracted from A. fumigatus, P. mar- (Qiagen, Hilden, Germany). For all amplifications a standard PCR neffei and T. stipitatus were then aligned and conserved regions program was used with the following cycle conditions: one dena- were targeted for primer design, using Primer3 (http://www.fok- turation step for 5 min at 95 °C, followed by 30 cycles with 30 s ker.wi.mit.edu/primer3/input.htm). The primer sets used for gene of denaturation at 95 °C, 30 s of primer annealing between 50 amplification and the expected sizes of the amplicons are shown and 60 °C depending on the TM of chosen primers, 30 s to 1 min in Table 1. of elongation at 72 °C following by a final extension of 7 min at We also amplified the ITS rDNA, partial beta tubulin and trans- 72 °C. For some primers, a touchdown annealing was used, from lation elongation factor genes (EF-1alpha), commonly used for 60 °Cto50°C for 30 s (0.5 °C per cycle) (see Table 1 for details). fungal phylogenies. We used respectively primers sets ITS 4/ITS Amplification products were separated on 1.5% agarose gels, stained 5(White et al., 1990), Bt2a/Bt2b (Glass and Donaldson, 1995) with SybrSafe (Biowhitaker Molecular Applications, Rockland, ME, and EF6/EF1D (Peterson et al., 2004). The sequences obtained USA) and photographed under UV illumination. PCR products were using the beta tubulin and EF-1 alpha genes could however not purified and sequenced at the Genoscope (Evry, France) in both be used because their introns were not alignable and because, directions using amplifying primers. Sequences were deposited at once they were removed, the sequences were not phylogeneti- GenBank under the numbers GU396463 to GU396534 for MS277, cally informative enough. We therefore, restricted the full analy- GU396381 to GU396462 for MS456, GU396604 to GU396667 for ses to these four DNA fragments: MS277, MS456, FG610 and the FG610, GU396535 to GU396603for ITS rDNA, GU454805 to ITS. GU454829 for MAT 1-1 and GU454830 to GU454847 for MAT 1-2 (all the alignments are available upon request). 2.3. Mating type gene amplification

2.5. Mating type detection and crosses under controlled conditions We attempted to amplify mating type genes using three different sets of primers. First we used the degenerated primers de- To test for sexual reproduction ability, complementary strains signed by Woo et al. (2006) to amplify mating type genes in from the putative asexual species P. pinophilum and P. funiculosum Penicillium species. These primers allowed the detection of mating previously identified as MAT 1-1 and MAT 1-2 were cultivated un- type genes in only a few strains for species phylogenetically close der controlled conditions. These species were chosen because they to P. marneffei. The sequences obtained using these primers were presented several strains of opposite mating types. For P. pinophi- aligned to develop a second set of primers (MAT 1-1-a and MAT lum crosses were performed using four isolates detected as MAT 1-2-a, Table 1). This second set of primers amplified fragments in 1-1 (strains MUCL 38548, IMI 211.742, LCP 1699 and LCP 1527) many strains, including those for which the first primers did not al- and three isolates detected as MAT 1-2 (strains CMPG 505, CMPG lowed amplification of mating type genes. However, the sizes of 568 and CMPG 1507). For P. funiculosum two MAT 1-1 isolates the amplicons obtained were small (<250 bp), impeding the analy- (LCP 3383 and CMPG 567) and two MAT 1-2 strains (LCP 3189 sis of selective pressures acting on mating type genes. We there- and CMPG 177) were used. Cultures were grown by pairs on fore developed a third set of primers (MAT 1-1-b and MAT 1-2-b, malt-agar plates under daylight at 25 °C. Microscopic observation Table 1) using the nucleotide sequences of the MAT 1-1 and MAT of plates were done after 15 days and then after 8 weeks of incuba- 1-2 idiomorphs from P. marneffei described by Woo et al. (2006) tion on unsealed plate. (GenBank Accession No. DQ340761 and DQ340762) and the MAT 1-1 and MAT 1-2 sequences from T. stipitatus (GenBank Accession No. EQ962658 and EQ962662 respectively). In all the cases, prim- 2.6. Phylogenetic analyses ers were designed in conserved regions of MAT 1-1 and MAT 1-2, using Primer3 online (http://www.frodo.wi.mit.edu/cgi-bin/pri For each isolate, sequence data obtained for both strands of mer3/primer3_www.cgi). each locus were edited and assembled using CodonCode aligner v

Table 1 Primers used for ampliﬁcation and sequencing.

Gene Primer Sequence (50–30) Annealing temperature (°C) Amplicon size ITS ITS-4 TCCTCCGCTTATTGATATGC 55 700 ITS-5 GGAAGTAAAAGTCGTAACAAGG Elongation Factor EF6 CTTSTYCCARCCCTTGTACCA 50 700 EF1d GGCCACGTCGATTCCGG Beta Tubulin BT 2 a GGTAACCAAATCGGTGCTGCTTTC 55 700 BT 2 b ACCCTCAGTGTAGTGACCCTTGGC MS277 ACACCYCACCARCAACTCAT Touchdown 60–50 770 ATCTGRAAGTCGCCCCATC M456 CTGATGGGTGATCCYGGTGT 50 500 TTGTTGTGCATGTGGACGTA FG610 CCGCAAYAAGATCGTCATCA 55 723 AGCATMTCCTGTGCATTCTTCA Mating type MAT 1-1a CGCCCTCTGAATAGTTTTATCG Touchdown 60–50 130 GCCCATTTTCCTTTGTAGGG Mating type MAT 1-2a AGGTTCCTCGACCCCCTAAT Touchdown 60–50 250 TTTTTCTCACACGGTTTGC Mating type MAT 1-1b CCACGTATAACGGGGCATC Touchdown 60–50 1000 CGGCTTGCCAMAGGTCTT Mating type MAT 1-2b GTGATAATGCTTSCGATAGAGAATG Touchdown 60–50 770 GTTGGAGAGGAGGCGTTGAC 696 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

1.3.4. (CodonCode Software). Initial nucleotide and corresponding 2.6.2.2. Comparison among individual gene trees. In order to use the amino acid alignments were made for individual genes using criterion of concordance between individual gene trees for species ClustalW (v 1.83) (Thompson et al., 1994) and T-coffee v 5.05 recognition in the Penicillium/Talaromyces group (Dettman et al., (Notredame et al., 2000) both with default settings. In order to 2003), we built restricted alignments that included only the strains be sure of the quality of the alignments and the reading frame, that could be amplified for the four DNA fragments (ITS, MS277, we aligned nucleotide sequences using the corresponding amino MS456 and FG610). A concatenated dataset using these four se- acid alignments as a guide. The reading frame of each nucleotide quences was created. Four single-gene trees and a concatenated tree sequence was determined using the emboss wise2 software and were constructed using ML. Following the criterion of concordance the guided alignment was done with the tranalign emboss pro- between multiple individual locus trees (Dettman et al., 2003; Le gram (http://www.emboss.sourceforge.net/). Manual edition of Gac et al., 2007), a clade was considered as an independent evolu- the alignments was done to remove gaps and verify their quality. tionary lineage (i.e. a ‘‘phylogenetic taxon”) when the clade was well Only unambiguously aligned regions were kept for further supported in at least two single-locus trees and was not contradicted analyses. by any other single-locus tree. A clade was considered as well supported when its basal node showed MP bootstrap proportions higher than 70%. To compare the individual gene trees we used the same in- 2.6.1. Phylogenetic reconstruction and comparison dexes as above (Felsenstein, 1989; Kuhner and Felsenstein, 1994; We used sequence alignments separately to obtain individual Nye et al., 2006; Robinson and Foulds, 1981). trees, under both likelihood and Bayesian frameworks. We tried The different full gene trees (including all strains yielding suc- to amplify the four different DNA fragments (ITS, MS277, MS456, cessful amplifications) were also compared by visual inspection, and FG610) in all the strains used in the study (Supplementary by checking whether differences in topologies were found at material). However, not all the sequences amplified in all the well-supported nodes. When discordant topologies were found, strains. Differences in the number of strains in each tree prevented tests were run to assess whether the differences were significant, formal comparisons between full individual trees. We therefore using the CONSEL package v 0.1i which implements the AU test created restricted overlapping datasets, including only the se- (Shimodaira and Hasegawa, 2001), the KH test (Kishino and Hase- quences of the strains that were successfully amplified for all the gawa, 1989), the SH test (Shimodaira and Hasegawa, 1999) and four sequences MS277, MS456, FG610 and the ITS. Using this re- computation of the RELL bootstrap proportions (Shimodaira and stricted dataset, we built (i) four single-sequence trees and (ii) a Hasegawa, 1999). These tests compare the p-value associated with single tree based on a concatenation of the sequences obtained each tree, which represents the possibility of that tree being the with a custom-made perl script. true tree given the data. The competing topologies are thus ranked according to their p-values in order to determine which one is the 2.6.1.1. Maximum likelihood analysis. PHYML (v 2.4.4) (Guindon and most likely. Gascuel, 2003) was used to infer phylogenetic trees from individual and concatenated alignments. Gaps were treated as missing 2.7. Determining selective pressure on mating type genes data. Modeltest v 3.7 (Posada and Crandall, 2001) was used to predict the best evolutionary model for tree inference. The chosen The evolution of mating type genes in sexual and asexual fungal model, based on the AIC and BIC ranking criteria in modeltest species is expected to differ with respect to the selective pressure was the GTR+I+G with 4 rate categories. The same model was used acting on each type of species. In sexual species there is pressure to for concatenated datasets. To determine node support for the tree maintain mating type genes functional, so strong purifying selec- topologies, bootstrap analysis was conducted using the consense tion is expected. On the contrary, asexual species should relax this program in the Phylip package (v 3.66) with 100 replicates and constraint and accumulate substitutions at a higher rate than sex- the majority consensus rule was chosen to obtain the final tree. ual species, which is called relaxed selection. Footprints of purify- For rooting the tree, the A. fumigatus sequence was chosen as ing selection may therefore be good evidence that sexual outgroup. reproduction is still important in nature or that it regularly occurred until recently. We were thus interested in characterizing the differences in selective pressure in the evolution of mating type 2.6.1.2. Bayesian analysis. MrBayes (v 3.1.2) was also used to infer genes in Penicillium vs. Talaromyces. We conducted tests on data- phylogenetic trees from individual alignments. The model chosen sets where all the available sequences of each species were in- with modeltest was GTR+I+G with 4 rate categories. Two indepen- cluded as well as on a subset of these datasets, including a single dent chains with 1 million generations each were conducted. The sequence per species, in order to focus on between-species substi- first 25% of the trees were discarded as burn-in and the remaining tutions. Powerful statistical methods are available to measure the trees were sampled every 100 trees for a total of 10,004 trees used selective constraint on genes, by calculating whether the genes to infer the consensus tree. Bayesian posterior probabilities were evolve neutrally, under purifying, or relaxed/positive selection obtained for the internal nodes. MrBayes determines which model based on the comparison between synonymous and nonsynony- suits each partition best. mous substitutions (these tests cannot distinguish between relaxed and positive selection as both increase the number of 2.6.2. Comparison among trees and use of concordance between nonsynonymous substitutions). multiple individual gene trees for species recognition We used the program codeml in the PAML 4 package (Yang, 2.6.2.1. Comparison among methods for each individual gene 1997, 2007) to measure the nonsynonymous/synonymous substi- tree. Maximum likelihood (ML) and Bayesian trees for each gene tution rate ratio (dN/dS), also referred to as x. x < 1 suggests puri- were compared by visual inspection by assessing whether differ- fying selection, x = 1 is consistent with neutral evolution, and ences in topologies could be observed at well-supported nodes. x > 1 is indicative of positive selection. x values close to or higher Further, we used two topological comparisons, the Robinson–Foul- than 1 may also indicate relaxed selection. Nested codon models ds (Robinson and Foulds, 1981) and the Kuhner–Felsenstein implementing the x ratio can be compared by means of a likeli- (Kuhner and Felsenstein, 1994) metrics, as implemented in the hood ratio test (LRT). We used the null model M1a, which assumes treedist program in phylip v 3.66 (Felsenstein, 1989), as well as two site classes with 1 > x > 0, and x1 = 1, which therefore implic- the Nye’s topological comparison (Nye et al., 2006). itly supposed that there are no sites under positive selection. This M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 697

model was compared with the alternative model M2a, which adds to sexuality) and qSA (transition from sexuality to asexuality), by an extra class of sites that allows x to take values higher than 1. the likelihood function that evaluates the data on a tree given We also compared the null model M7, which assumes a beta distri- the rates, by the prior probability distributions for the rate param- bution of x across sites, with the alternative model M8, which eters, and by the set of 100 trees as obtained from a bootstrap anal- adds an extra class of sites to M7 where x can take values higher ysis performed by PHYML on the aligned sequences. than 1. Thereby positive selection can be detected if a model allow- In our analysis, we constrained the rate qAS to 0, as a transition ing for positive selection is significantly more likely (as estimated from asexuality to sexuality is highly unlikely. We used uniform by the LRT) than a null model without positive selection. We prior distributions for the other parameters, assuming that all val- checked for significant variability across sites by means of a LRT ues of the parameters were equally likely a priori. The principle of that compares model M0, which assumes a single x value for all the Bayesian estimation of the parameters is as follows: at each sites in the alignment, with model M3, which assumes that x is iteration the chain proposes a new combination of rate parameters distributed in three classes of site, x0 =0,x1 = 1, and x2 that can and randomly selects a new tree from the sample of 100 initial take values higher than 1, thus allowing for the detection of posi- ones. The likelihood of the new combination is evaluated and this tive selection. new state of the chain is accepted or rejected (for more details, see We were also interested in contrasting the selective pressures Pagel et al., 2004). We ran the Markov chain for 5,050,000 itera- acting in the mating type genes vs. the house-keeping genes in or- tions, discarded the 50,000 first sets of parameters and sampled der to test: (i) whether mating type genes are under different from the chain every 100 generations. This yielded 50,000 sets of selective constraints than the house-keeping genes employed in rate coefficients from which we estimated the joint probability dis- this study for building the phylogeny and (ii) whether the mating tribution. Each sampled set of coefficient was also used to recon- type genes are less constrained (i.e., have higher substitution rates) struct ancestral states at each node of the majority-rule in asexual species than in sexual ones. To do so, we used the consensus tree obtained from the 100 initial trees. Note that the parameter estimates of models M1a and M2a to determine the transition rate from asexuality to sexuality was set to 0, each node fraction of sites that belonged to each class of sites according to leading to both sexual and asexual species having a probability of their x values (i.e., 1 > x >0,x1 = 1 and x2 > 1). being sexual equals to 1. To take into account phylogenetic uncer- Selective pressure is expected to vary along the different lin- tainty, we used the most recent common ancestor (MRCA) ap- eages in the phylogeny, as some species may be under higher or proach proposed by Pagel et al. (2004). For each node in the lower constraints relative to the overall pressure across the tree. consensus tree, we identified in each tree the MRCA of the group

We explored dN/dS variability along the different branches of of species and reconstructed the state at that node and combined MAT and marker gene trees by assuming an independent x ratio the information at that node. for each lineage. This corresponds to the ‘‘free-ratios” test, which can be compared to model M0 (where all branches are assumed 3. Results to have the same x value) by a LRT. This is not a formal test to test for branch-specific variations in dN/dS ratio (for branch-specific 3.1. Phylogenetic analysis of single-locus datasets and species models see Yang, 1998; Yang and Nielsen, 1998), but we used it recognition as an exploratory means to detect species of interest that could be further validated by other tests and also experimentally. 3.1.1. Comparison between methods of reconstruction for each Finally, for specific branches of interest, we used branch-site individual gene tree models (Yang and Nielsen, 2002; Zhang et al., 2005). These are Taxonomic identifications of the isolates were validated by powerful methods that can predict specific sites at particular comparison of their ITS sequences with those of ex-type or repre- branches to evolve under higher substitution rates, as they allow sentative isolates (Lobuglio et al., 1993) deposited in GenBank. x to vary at different branches in the tree. In branch-site models, Using maximum likelihood and Bayesian analyses, we obtained it is assumed that branches are divided a priori into foreground individual gene trees for the ITS, MS277, MS456, and FG610 genes. and background lineages, and positive selection is allowed to occur The tree generated using MS277 was both the one with the highest only at foreground branches. We used alternative model A that as- number of strains and the best resolved of all the trees produced sumes four classes of sites: class 0 includes the codons that are (Fig. 1). All other trees are presented in Supplementary material conserved throughout the tree (1 > x > 0). Site class 1 includes co- (Figs. S1–S3). Single-gene trees resulting from Bayesian and ML dons that are neutral across the entire phylogeny (x1 = 1). Site analyses were compared using three different metrics, Robinson– classes 2a and 2b include codons that are conserved or neutral Foulds (Robinson and Foulds, 1981), Kuhner–Felsenstein (Kuhner on background branches but can have high substitution rates on and Felsenstein, 1994) and Nye’s topological comparison (Nye foreground branches (x2 > 1). As a null model we used model A, et al., 2006). Table 2 shows the results of the comparisons. All but with x2 fixed to 1 (x2 = 1). We implemented a LRT that com- topologies were highly congruent according to the three metrics pares null model A with alternative model A in order to test for employed, between ML and Bayesian trees. point events of high substitution rates.

2.8. Ancestral state reconstruction 3.1.2. Phylogenetic relationships within Penicillium subgenus Biverticillium To estimate the number of events of putative sex losses in the Three major robust monophyletic clades (>98% Bootstrap Pro- Penicillium/Talaromyces group, we performed an ancestral state portions) were reconstructed within the subgenus Biverticillium, reconstruction analysis, accounting for phylogenetic uncertainty among species investigated in this study (Fig. 1). All the terminal (Pagel et al., 2004). We use an alignment of the locus MS277 con- clades representative of the species were also significantly sup- taining a single individual per species, chosen at random, and using ported, with the exception of P. funiculosum. A main clade groups A. fumigatus as a sexual outgroup. We constructed a Markov chain most of the isolates of P. funiculosum and includes P. rubrum to implement a model of trait evolution, and used the chain to esti- isolates, while the neotype strain of P. funiculosum (MUCL mate the posterior probability distribution of the rate coefficient 38969) groups outside (73% BP) with two Talaromyces sp. iso- and of the nature of the ancestral states at each node. The model lates (LCP 4272 and LCP 4224). Talaromyces and Penicillium spe- in our case is defined by two rates, qAS (transition from asexuality cies are interspersed in the phylogeny. The clade I supports the 698 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

Talaromyces macrosporus MAT 1-1 100 100 T. flavus MAT 1-2 100 100 96 P. aculeatum

99 P. funiculosum 65

100 100 P. rubrum 68 100 P. funiculosum 96 73 Talaromyces sp P. funiculosum 91

P. pinophilum 90 100 72

100 67 94 Clade III 88 100 P. purpurogenum var. rubrisclerotium P. verruculosum P. marneffei 100 T. stipitatus 97 100 P. erythromellis 66 88 100 P. minioluteum 100 97 99 T. udagawae 81 Clade II 73 100 T. trachyspermus 100 100 98 100

P. purpurogenum 100 P. piceum P. loliense 77 Talaromyces sp 100 100 91 P. variabile 89 90 T. wortmanii 98 P. loliense Clade I 100 65 P. rugulosum 100 T. bacillisporus

Fig. 1. Tree based on the gene MS277 (ribosome biogenesis protein) generated by maximum likelihood using the model GTR+I+G with 4 rate categories and 100 bootstrap replicates. Bootstrap values higher than 65% are reported at each node. Penicillium species are putatively asexual while Talaromyces species have a sexual stage described. Strains where mating type genes could be amplified are indicated (MAT 1-1 genes white circles; MAT 1-2 genes full circles). Only strains for which the gene MS277 was amplified are shown in the figure, so that some strains used in the study are lacking in the tree. The scale indicates the number of expected nucleotide substitutions along the branch. M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 699

Table 2 evolves under different rates in the two misplaced species than Pairwise topological comparisons between Maximum Likelihood and Bayesian in the other species, possibly being less constrained. individual gene trees for each gene using three different metrics. These full gene trees include all strains yielding successful amplifications. 3.1.4. Species recognition: comparison between individual gene trees Gene Robinson– Nye’s topological score Kuhner– built using the restricted dataset Foulds (%) Felsenstein In order to use the criterion of concordance between multiple ITS 56 84.2 1.509272e01 gene genealogies to delimit species, we compared the trees gener- MS277 32 93.9 5.592012e01 MS456 59 85.4 3.077379e01 ated by maximum likelihood using four independent nuclear loci FG610 38 87.3 9.223337e01 (ITS, MS277, MS456 and FG610) on the exact same set of strains, those for which we could obtain sequences for all the four DNA fragments. A tree was also built using a concatenation of these alignments (Fig. 2). Bold branches on this concatenated tree indi- relationships between T. bacillisporus and T. wortmanii with P. cate the clades identified as independent evolutionary lineages. loliense, P. piceum, P. rugulosum and P. variabile. The clades II The morphological species delimitation was supported by the cri- and III are sister clades. In the clade II, the relationships between terion of concordance between multiple gene genealogies for all T. trachyspermus and P. purpurogenum and between T. udagawae the six species represented by at least two strains in the restricted and P. minioluteum are well supported (P97% BP), while P. ery- dataset (P. funiculosum, P. pinophilum, T. stipitatus, T. trachyspermus, thromellis is not significantly grouped with the two latest. In P. minioluteum, and P. piceum). The phylogenetic relationships were the clade III, T. stipitatus is in a basal position to a major robust in agreement with the relationships previously inferred based on sub-clade containing T. flavus, T. macrosporus, P. pinophilum, P. morphological and physiological characters (Pitt, 1979). The com- funiculosum; P. rubrum, P. marneffei, P. aculeatum, P. purpuroge- parisons of the individual gene trees using the Robinson–Foulds’ num var. rubrisclerotium and P. verruculosum. However relation- and Kuhner–Felsenstein’s metrics, and the Nye’s topological com- ships within this sub-clade are not resolved. parison are given in Table 4. All topologies were overall congruent When comparing our results to LoBuglio et al. (1993), Talaromy- according to the three metrics employed. ces and Penicillium species groupings were maintained even if species sampling was not exactly the same. In our study bootstrap 3.2. Phylogenetic distribution of sexuality and reconstruction of the values have reached 97% for clade II as a sister of clade III, and ancestral state of the mode of reproduction 99–100% for each clade. The sexual and putatively asexual clades are interspersed in the 3.1.3. Comparison between individual gene trees built using the full phylogeny, suggesting that there have been several independent datasets losses of the sexuality in the fungal subgenus Penicillium Biverticil- When comparing by visual inspection the individual gene trees lium. The ancestral character state reconstruction in fact inferred a built using the full sequence datasets, the placements of a few total of 10 putative sex losses (Fig. 3), considering that all the puta- strains were different between some of the trees, but the corre- tively asexual species are really so. sponding nodes had so little support that these differences were not meaningful. The only incongruence that concerned well-sup- 3.3. Phylogenetic distribution of mating types and selective pressure on ported nodes was the one involving the two T. trachyspermus mating type genes strains CMPG 545 and CMPG 563, that had different placements in the ITS tree as compared to all the other trees. Using the en- The first steps in our study for approaching the reproductive forced topology test, we tested whether the ITS data significantly mode of putatively asexual species of Penicillium was to search supported a topology that differed relative to the other single-gene for mating type genes by PCR and to assess the selective pressure trees. We therefore constrained the ITS tree to place the two prob- they were subject to. We used three different sets of primers to at- lematic strains as they appear in all other trees. The CONSEL anal- tempt amplifications of mating type genes in all the species used in ysis was significant, suggesting that the topology supported by the this study. MAT genes were amplified in several species of Penicil- ITS data is in fact significantly different from the other single gene lium and Talaromyces mainly from the species phylogenetically topologies (Table 3). The two misplaced T. trachyspermus strains close to P. marneffei and T. stipitatus, that were the species used (CMPG 545 and CMPG 563) have unexpectedly long branches to design the mating type primers (Table 5). In some species, (0.084 and 0.125 in the ML and Bayesian trees, respectively) as MAT 1-1 genes presented a small intron (20 bp) while all MAT 1- compared to the clade where they should be placed, which has a 2 sequences presented one putative conserved intron (50 bp) (Data branch length of 0.010 in the two analyses. This suggests that ITS not shown). All introns were removed for further analyses. A single

Table 3 CONSEL analysis for the enforced topology tests.

Tree Obs au np bp pp kh sh wkh wsh

ITS 2.0 0.679 0.690 0.693 0.882 0.694 0.694 0.694 0.694 Modiﬁed ITS 2.0 0.321 0.310 0.307 0.118 0.306 0.306 0.306 0.306 au p-values for the approximately unbiased (AU) test. np bootstrap probability of the selection. bp same as np, but calculated directly from the replicates. pp Bayesian posterior probability (pp) calculated by the BIC criterion. kh p-values for the Kishino–Hasegawa (KH) test. sh p-values for the Shimodaira–Hasegawa (SH) test. wkh p-values for the weighted Kishino–Hasegawa (WKH) test. wsh p-values for the weighted Shimodaira–Hasegawa (WKH) test. For more details see Shimodaira and Hasegawa (2001). 700 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

Fig. 2. Tree generated by maximum likelihood analysis based on four concatenated nuclear loci (ITS, MS277, MS456 and FG610). Bold branches indicate the clades identiﬁed as independent evolutionary lineages supported by the majority of the loci. Bootstrap values higher than 70% are reported at each node.

Table 4 Pairwise topological comparisons, using three different metrics (A, Robinson–Foulds’distance, B, Kuhner–Felsenstein’distance and C, Nye’s topological score), between individual gene trees and the concatenated tree, using the restricted dataset including only the strains for which all four genes could be sequenced.

MS277 ITS MS456 FG610 Concatenated dataset A MS277 0 32 32 26 20 ITS 32 0 32 34 32 MS456 32 32 0 32 26 FG610 26 34 32 0 18 Concatenated dataset 20 32 26 18 0 B MS277 0 361.335 369.801 365.508 307.919 ITS 361.335 0 355.626 358.398 375.77 MS456 369.801 355.626 0 356.7 324.877 FG610 365.508 358.398 356.7 0 300.978 Concatenated dataset 307.919 375.77 324.877 300.978 0 C MS277 0 74.1% 74% 76.5% 84.2% ITS 0 72.6% 71.7% 73.4% MS456 0 75.4% 78.8% FG610 0 86% Concatenated dataset 0 partial mating type gene could be ampliﬁed in several Penicillium they were expected to carry both MAT idiomorphs. However, the species, as well as in the strains from T. trachyspermus, T. wortmanii phylogenetic distance of these species from T. stipitatus and P. mar- and T. udagawae. These Talaromyces species being homothallic, neffei could have impeded annealing of the primers. In contrast, M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 701

Fig. 3. Ancestral character state reconstruction of sexuality and asexuality based on the MS277 gene using the Bayesian estimation method. Light gray boxes represent sexual species. Black boxes represent asexual ones. At each node, the filled circle represents the probability that this node is at a particular state: light gray for sexual, dark for asexual. Note that the probability that a given node is considered as sexual is equal to 1 as long as one of its descendants is asexual. This is due to the constraint we applied on the analysis that disables transitions from asexuality to sexuality (see text). two mating type genes were detected in the six strains of T. flavus Analyzed together, the MAT 1-1 sequences showed 72% of fairly and in the two strains of T. stipitatus used in this study (Fig. 1). Mat- conserved sites, at x = 0.14 (Fig. 4). When analyzed separately, ing type genes could thus be amplified in the putatively asexual MAT 1-1 sequences in Penicillium (the ‘‘asexual” group) showed species close to P. marneffei and T. stipitatus (Table 5, Fig. 1). In P. 85% of conserved sites, with x = 0.17, whereas the MAT 1-1 se- pinophilum, MAT 1-1 was detected in 5 out of 14 strains while quences in Talaromyces (the sexual group) appeared slightly less MAT 1-2 was detected in 9 out of 14 strains. In P. funiculosum, most conserved, with x = 0.20 at 76% of sites. We had not enough of the strains were detected as MAT 1-1, MAT 1-2 being amplified MAT 1-2 sequences from Talaromyces to estimate x separately in only 2 out of 14 strains. In P. erythromellis, all the four analyzed but the overall estimated value was x = 0.22 at 93% of sites. From strains seemed to carry only MAT 1-1 (Table 5, Fig. 1). these results, MAT 1-1 seems to be under stronger purifying selec- Overall, most Talaromyces strains showed amplification of both tion, but at a smaller fraction of sites than MAT 1-2 (x = 0.14 in MAT 1-1 and MAT 1-2 sequences, in agreement with the reports of 72% of sites vs x = 0.22 in 93% of sites). Overall, MAT 1-1 and homothallism in these species (Takada and Udagawa, 1988). In con- MAT 1-2 genes in all the analyzed species were highly conserved, trast, although all the strains of Penicillium were also systematically even in species considered as asexual. The results from the screened for MAT 1-1 and MAT 1-2 sequences, only a single mating branch-site analysis however suggest that there are two sites with type was detected in all the Penicillium strains, which would corre- high substitution rates (x2 = 41.23) in MAT 1-1 in the lineage lead- spond to a heterothallic breeding system in sexual species. ing to the species P. pinophilum (CMPG 505, LCP 3569 and LCP 2604) and P. purpurogenum var. rubrisclerotium (MUCL 29225), 3.4. Selection pressures on MAT genes i.e. to a Penicillium clade, putatively asexual. We ran the same analyses as above using a dataset including a No evidence of positive or relaxed selection was found in any of single strain per species to concentrate on between-species substi- the tests performed using all the strains (i.e. multiple strains per tution rates, and we obtained similar results (Table 6). As above, no species). Instead, both MAT 1-1 and MAT 1-2 genes had low dN/ positive or relaxed selection was detected, except for the MAT 1-1 dS values in all the models used. Likelihood ratio tests comparing gene that showed two sites as having increased substitution rates respectively models M1a vs. M2a, and models M7 vs. M8, rejected at a particular branch leading to P. pinophilum (x2 = 34.64). the models that allow for increased substitution rates in favor of House-keeping genes were also under strong purifying selection the null model where dN/dS values > 1 are not permitted (Table 6). (analyses using the full dataset), and appeared even more con- Furthermore, most of the sites in MAT genes showed x values strained (Table 6): MS277 had 72% of sites with x = 0.14; MS456 well below one. Table 5 and Fig. 4 show the LRTs for models M1a showed the strongest purifying selection with x = 0.007 at 94% and M2a for the two MAT genes as well as for the house-keeping of the sites; FG501 had 95% of sites at x = 0.03, and FG610 exhib- genes, using the full dataset, i.e. multiple strains per species. ited 94% of sites under x = 0.02 (Fig. 4). 702 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

Table 5 Table 6 Strains where mating type genes were detected. LRTs for site-speciﬁc, free-ratios and branch-site analyses for MAT genes for all strains and one strain per species. Collection no. Mating type detected Species All strains LCP 2604 MAT 1-1 Penicillium sp. LRT 2d Degrees of freedom p-value CMPG 568 MAT 1-1 P. pinophilum CMPG 505 MAT 1-1 P. pinophilum MAT 1 all strains MUCL 46114 MAT 1-1 P. pinophilum M0 vs. M3 52.6 2 3.78491E12 CMPG 1507 MAT 1-1 P. pinophilum M1a vs. M2a 0 2 1 LCP 3569 MAT 1-1 P. pinophilum M7 vs. M8 0.5 2 0.778022372 MUCL 14090 MAT 1-2 P. pinophilum M0 vs. FR 55.45 13 3.36849E07

LCP 3438 MAT 1-2 P. pinophilum Manull vs. MAalt 14.74 1 0.0001234 LCP 5166 MAT 1-2 P. pinophilum One strain/species MUCL 38548 MAT 1-2 P. pinophilum M0 vs. M3 43.5 2 3.58175E10 LCP 1699 MAT 1-2 P. pinophilum M1a vs. M2a 0 2 1 LCP 1527 MAT 1-2 P. pinophilum M7 vs. M8 0 2 1 IMI 211.742 MAT 1-2 P. pinophilum M0 vs. FR 54.28 26 0.000935868 MUCL 47343 MAT 1-2 P. pinophilum Man vs. MA 13.19 1 0.000281447 MUCL 19010 MAT 1-2 P. pinophilum ull alt Manull vs. MAalt 5.37 1 0.02048598 MUCL 29225 MAT 1-1 P. purpurogenum MAT 2 all strains var. rubrisclerotium M0 vs. M3 5.5 2 0.063927861 FRR 1868 MAT 1-1 P. erythromellis M1a vs. M2a 0 2 1 LCP 2233 MAT 1-1 P. erythromellis M7 vs. M8 0 2 1 LCP 3727 MAT 1-1 P. erythromellis M0 vs. FR 6.05 6 0.417612437

LCP 4464 MAT 1-1 P. erythromellis Manull vs. MAalt 01 1 LCP 3383 MAT 1-1 P. funiculosum Markers MS277 CMPG 567 MAT 1-1 P. funiculosum M0 vs. M3 117.61 2 2.89276E26 CMPG 921 MAT 1-1 P. funiculosum M1a vs. M2a 0 2 1 CMPG 534 MAT 1-1 P. funiculosum M7 vs. M8 0.16 2 0.923116346 CMPG 233 MAT 1-1 P. funiculosum MS456 LCP 5346 MAT 1-1 P. funiculosum M0 vs. M3 95.18 2 2.14746E21 CMPG 811 MAT 1-1 P. funiculosum M1a vs. M2a 0 2 1 CMPG 556 MAT 1-1 P. funiculosum M7 vs. M8 0.54 2 0.763379495 CMPG 757 MAT 1-1 P. funiculosum CMPG 527 MAT 1-1 P. funiculosum FG610 CMPG 34 MAT 1-1 P. funiculosum M0 vs. M3 108.68 2 2.51442E24 LCP.5345 MAT 1-1 P. funiculosum M1a vs. M2a 0 2 1 CMPG 177 MAT 1-2 P. funiculosum M7 vs. M8 1.36 2 0.506616995 LCP 3189 MAT 1-2 P. funiculosum LCP 4991 MAT 1-1 P. rubrum controlled conditions. Sexual structures and ascospores were ob- MUCL 38781 MAT 1-1 P. verruculosum served in controlled conditions for the cross of two strains belong- MUCL 38803 MAT 1-1 P. aculeatum ing to species P. pinophilum (strains CMPG 505 MUCL 38548) LCP 2692 MAT 1-1 P. aculeatum (Supplementary material). These two strains of opposite mating LCP 2466 MAT 1-2 P. aculeatum types placed under controlled conditions produced ascogenous hy- LCP 2892 MAT 1-1/MAT 1-2 T. flavus phae and antheridia after 15 days and cleistotothecia after 30– LCP 3067 MAT 1-1/MAT 1-2 T. flavus 45 days. Microscopic observation of the crosses was made using LCP.2888 MAT 1-1/MAT 1-2 T. flavus methylene blue/lactophenol staining to contrast the structures LCP 2885 MAT 1-1/MAT 1-2 T. flavus LCP 2481 MAT 1-1/MAT 1-2 T. flavus that are hyaline. Conidiophores and conidia were produced 6 days LCP 4144 MAT 1-1 /MAT 1-2 T. flavus after incubation, but penicilli were difficult to observe when cul- LCP 4437 MAT 1-1/MAT 1-2 T. macrosporus. tures became old. To facilitate the observation of ascogonia and LCP 4440 MAT 1-1/MAT 1-2 T. macrosporus antheridia we took samples at the periphery of the colony, where LCP 1489 MAT 1-1 T. wortmanii conidiophores are rare (Supplementary material, Fig. S7). Cleisto- LCP 4224 MAT 1-1 /MAT 1-2 Talaromyces sp. thecia were formed but were long to ripe. Ascospores were ob- LCP 4272 MAT 1-1/MAT 1-2 Talaromyces sp. served only when media dried (90 days). In P. funiculosum,no LCP 4434 MAT 1-1 Talaromyces sp. evidence of sexual development was detected. LCP 4435 MAT 1-1 Talaromyces sp. CMPG 545 MAT 1-1 T. trachyspermus CMPG 563 MAT 1-1 T. trachyspermus LCP 2889 MAT 1-1 T. trachyspermus 4. Discussion LCP 2890 MAT 1-1 T. trachyspermus LCP 2884 MAT 1-1 T. trachyspermus Multiple gene genealogies were used in this study to compare CMPG 1216 MAT 1-1 T. udagawae phylogenetic and morphological criteria for species recognition LCP 4441 MAT 1-1/MAT 1-2 T. stipitatus and to infer phylogenetic relationships among the most common LCP 4439 MAT 1-1/MAT 1-2 T. stipitatus species of the Penicillium/Talaromyces group. The individual gene trees (ITS, MS456, MS277 and FG610) showed overall concordant topologies and we could obtain well resolved phylogenies. The species were grouped into three major clades, in agreement with 3.5. Inducing sex in Penicillium species under controlled conditions morphology. Clade I contains species with yellow mycelium that grow slowly on the media recommended for the taxonomy of the To test for sexual reproduction ability, complementary strains group (Pitt, 1979). The sister clades II and III are more diverse from the ‘‘asexual” species P. pinophilum and P. funiculosum previ- and contain species producing red pigmentation. Pitt (1979) used ously detected as MAT 1-1 and MAT 1-2 were cultivated under growth rates as the primary distinctive character to accommodate M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 703

Fig. 4. Histograms indicating the number of sites falling in different classes of omega values (dN/dS) for the MAT 1 (A) and MAT 2 (B) genes in the analysis using the full dataset (i.e., multiple strains per species). non synnematous species of the subgenus Biverticillium into the The genes used to build the phylogenies in this study showed a Series Miniolutea and Islandica of the Section Simplicium. Here we good power for solving the relationships between the different have shown a clear overlap between the Ser. Islandica and the species. The MS277 tree provided significantly higher resolution Clade I. The previous uncertainty about the placement of P. erythro- and better support than the rest of the genes used, including the mellis (only known then from the Type isolate in 1979) in Ser. Islan- ITS. Two of the most commonly used genes for fungal phylogenies, dica, slow growing but showing red pigmentation more akin to the EF-1alpha and beta-tubulin genes, did not provide useful infor- some members of the Ser. Miniolutea, is resolved in our molecular mation. Altogether these results are in accordance with the study phylogeny in favor to Ser. Miniolutea. The species of the Series of Aguileta et al. (2008) and Marthey et al. (2008) in showing that Miniolutea segregate into two sister clades: clade II, composed of MS277 appears to be one of the most powerful genes for fungal P. erythromellis, P. purpurogenum and P. minioluteum, and clade phylogenies. III, including some species of the complex P. funiculosum (P. funicu- Beyond a better understanding of the phylogenetic relation- losum, P. pinophilum, P. purpurogenum var. sclerotium and P. rubrum) ships in the Penicillium/Talaromyces group, we were interested with the closely related P. aculeatum and P. verruculosum. All the in reconstructing the evolutionary history of the reproduction species analyzed in this study are phylogenetically well delimitat- mode in this genus. We thus looked at the distribution of sexual ed, with the exception of the taxonomically confused complex P. Talaromyces and putative asexual Penicillium species in the tree. funiculosum (Van Reenen-Hoekstra et al., 1990), including associ- Penicillium species were interspersed in the phylogeny among ated Talaromyces. However, we have shown here a clear phyloge- Talaromyces species, confirming that they do not represent a sin- netic delimitation of P. minioluteum and P. pinophilum, one time gle clade, as previously shown (Lobuglio et al., 1993). Morpho- synonymised with P. funiculosum (Raper and Thom, 1949). logical characters used in the classification schemes to separate 704 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 taxonomically Penicillium and Talaromyces species (Pitt, 1979) did This idea of recent loss of sex is also supported by the evidence not reflect phylogenetic relationships, but only the ability of of high substitution rates in the MAT 1-1 gene in the Penicillium in vitro induction of sexual structures. Considering that the puta- clade containing the species P. pinophilum. Such a high substitution tively asexual species are indeed purely clonal, we inferred 10 rate can indeed be due to relaxed selection. It could also be due to independent losses of sex in the Penicillium/Talaromyces group. recruitment of the gene in other functions than mating. Wik et al. We have used some additional species compared to Lobuglio (2008) reported that mating type genes in Neurospora evolved fas- et al. (1993), which reveal further events of sex losses. These ter than several nuclear genes and that they were under positive findings are relevant to the study of the maintenance of sex de- selection even in heterothallic species. This has been explained spite its cost, across the tree of life (Otto and Lenormand, 2002). by the fact that, in Neurospora, mating type genes are also involved It reinforces the idea that sex is often lost, but asexuality re- in vegetative incompatibility. Vegetative incompatibility genes mains in terminal branches, because these asexual clades are may indeed be highly variable and sometimes evolve under diver- not able to persist long enough to diversify (Otto and Lenor- sifying selection because of frequency dependent selection (Wu mand, 2002). et al., 1998). Other functions have also been attributed to mating Because such conclusion relies on the assumption that the puta- types genes. In Saccharomyces cerevisiae, mating type genes take tively asexual species are effectively so, we attempted to detect part in maintaining cell wall integrity and stress response (Verna mating type genes in these species and to analyze the selective and Ballester, 1999). Also in S. cerevisiae, the simultaneous expres- pressure they were subjected to. We were able to detect sequences sion of both mating-type alleles in haploid yeasts maintains the of mating type genes in several species, including Penicillium spe- stability of the microtubules involved in mitosis processes (Stein- cies. More generally, mating type genes have been found in most berg-Neifach and Eshel, 2000). of the fungal species where they have been searched, regardless In the group Pencillium/Talaromyces, all the species already of their sexual and asexual status. Only in one (sexual) species, known to reproduce sexually are homothallic except the hetero- the ascomycete yeast Lodderomyces elongisporus, mating type thallic T. derxii (Takada and Udagawa, 1988). The amplification of genes have been reported to be absent after genome sequencing both MAT 1-1 and MAT 1-2 in several Talaromyces species is in (Lin and Heitman, 2007). agreement with homothallism. In contrast, a single mating type Mating type genes in our study were found to be highly con- idiomorph was detected in all the Penicillium species analyzed in served, even in the putatively asexual Penicillium species, although this study, suggesting that, if those species reproduce sexually, his substitution rates were detected in a Penicillium clade. Mating they are heterothallic. If they are in fact asexual and the lack of type genes seemed to be under weaker purifying selection than amplification is not merely due to mutations in the primer regions the house-keeping genes, but this was true both in sexual or (which could be likely under relaxed selection in asexual species), asexual fungi. These results are also consistent with other studies the results indicate that the Penicillium species had heterothallic in fungi: in almost all the putatively asexual species where mating sexual ancestors, or alternatively, they have lost one mating type type genes have been sequenced, they presented no evidence of after having lost sexual ability. The distribution of the MAT 1-1 loss-of-function mutations (O’Donnell et al., 2004). Only in the and MAT 1-2 sequences in the Clade III, with basal clades present- entomopathogenic fungus Cordyceps takaomontana did mating ing both idiomorphs, may suggest that homothallism is ancestral type loci contain pseudogenes (Yokoyama et al., 2003). This sug- in this group, but additional data would be needed to reconstruct gests that in most fungi, mating type genes may still be functional, the ancestral state. In fungal groups such as Cochliobolus and which is not expected if these species were truly asexual for long. Fusarium, homothallism was suggested to be derived from hetero- However, mating type genes may also be conserved in asexual spe- thallism several times independently by fusion or linkage of the cies if they are involved in other functions, which has in fact been two idiomorphs (O’Donnell et al., 2004; Yun et al., 1999). In other reported (Wik et al., 2008; Verna and Ballester, 1999; Steinberg- groups, such as Aspergillus, the reverse situation seems to have Neifach and Eshel, 2000). occurred. Heterothallic species may appear from a hypothetical The finding that MAT genes in Penicillium species seemed to homothallic ancestor presenting closely linked mating type genes mostly evolve under purifying selection however raises the ques- separated by gene translocation and with subsequent gene loss tion of whether they are able to undergo sexual reproduction. In (Galagan et al., 2005). the species P. pinophilum, strains collected in several continents The evolution of homothallism and heterothallism is interesting showed relatively balanced frequencies of MAT 1-1 and MAT 1-2 from an evolutionary point of view (Billiard et al., in press). Hetero- strains, which is an indication of the occurrence of more or less re- thallism has traditionally been considered as a mechanism that cent sex events. Further, these P. pinophilum strains carrying MAT promotes outcrossing, while homothallism would favor selfing. 1-1 and MAT 1-2 are intermingled in the phylogeny, suggesting However, diploids of heterothallic fungi are heterozygous at the that they recombine in nature, or at least that recombination oc- mating-type locus and meiosis segregates haploid genotypes that curred until recently. Strains with opposite mating types were in are compatible for mating and are thus able to undergo selfing. fact able to initiate sexual reproduction and clesitothecia and asci In the case of homothallism, mating with an identical haploid production. Ascospore formation took significantly longer than for has little advantage over asexual reproduction, being unable to cre- the sexual Talaromyces, but our laboratory conditions may not be ate any genetic variation by recombination, or to repair DNA. optimal to trigger its sexual reproduction. Some species considered Therefore, it has been suggested that homothallism has instead as asexual, like A. fumigatus, have previously been found to repro- evolve in outcrossing species to allow universal compatibility duce sexually under controlled conditions, but more slowly than among gametes (Billiard et al., in press; Giraud et al., 2008), while their sexual relatives and only after several months of culturing heterothallism would have evolved to avoid mating between iden- (O’Gorman et al., 2009). Here, we were not able to observe mature tical haploids (Czaran and Hoekstra, 2004). Interestingly, some ascospores and the following crossing experiments produced homothallic fungi, such as the homothallic Neurospora species, rely immature cleistothecia. This prevented testing ascospore viability mainly on the intra-haploid mating for reproduction and neither to study the fitness consequences of sexual reproduction and to asexual conidia nor outcrossing have been observed in nature perform analyses of the progeny in order to ensure that the ascosp- (Glass et al., 2000). In this case, sexual reproduction by intra-hap- ores were in fact produced after meiosis and recombination. Alto- loid mating could be selected for if it provides advantages not re- gether, these results suggest that P. pinophilum may have lost sex lated with recombination. In fungi, such advantages include only recently. production of resistant sexual spores, cell rejuvenation and senes- M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706 705 cence avoidance and reduced accumulation of slightly deleterious Coppin, E., Debuchy, R., Arnaise, S., Picard, M., 1997. Mating types and sexual mutations (Aanen and Hoekstra, 2007; Bruggeman et al., 2003; development in filamentous ascomycetes. Microbiol. Mol. Biol. Rev. 61, 411– 428. Haedens et al., 2005). Couch, B.C., Fudal, I., Lebrun, M.H., Tharreau, D., Valent, B., Van Kim, P., Notteghem, Regarding sexual ability in Penicillium species, we could obtain J.-L., Kohn, L.M., 2005. Origins of host-specific populations of the blast pathogen the first steps of sexual reproduction in P. pinophilum, but we could Magnaporthe oryzae in crop domestication with subsequent expansion of pandemic clones on rice and weeds of rice. Genetics 170, 613–630. not obtain mature cleistothecia. Therefore, there is no compelling Czaran, T.L., Hoekstra, R.F., 2004. Evolution of sexual asymmetry. BMC Evol. Biol. 4, 34. evidence of efficient sexual reproduction, although sexuality in this De Visser, J.A., Elena, S.F., 2007. The evolution of sex: empirical insights into the species cannot be completely ruled out. We detected mating type roles of epistasis and drift. Nat. Rev. Genet. 8, 139–149. Dettman, J., Jacobson, D., Taylor, J., 2003. A multilocus genealogical approach to genes that appeared to be still functional, based on their sequences phylogenetic species recognition in the model eukaryote Neurospora. Evolution, and the ability of the strains to undergo at least the first steps of 2703–2720. sexual reproduction, although they appeared to evolve more rap- Devier, B., Aguileta, G., Hood, M.E., Giraud, T., 2009. Ancient trans-specific polymorphism at pheromone receptor genes in Basidiomycetes. Genetics 181, idly than in sexual species, in particular MAT 1-1. The only way 209–223. to confidently elucidate the mode of reproduction of the fungus Domsch, K.H., Gams, W., Anderson, T.H., 1980. Compendium of Soil Fungi. Academic in nature is to sample large populations and look for footprints of Press, London. recombination (Taylor et al., 1999), but we could not obtain en- Dyer, P.S., Paoletti, M., 2005. Reproduction in Aspergillus fumigatus: sexuality in a supposedly asexual species? Med. Mycol. 43, S7–S14. ough strains of each species for such studies. Felsenstein, J., 1989. PHYLIP – phylogeny inference package (version 3.2). Cladistics In conclusion, our study provides a robust and comprehensive 5, 164–166. phylogeny of the Penicillium subgenus Biverticillium, improving Fisher, M.C., 2007. The evolutionary implications of an asexual lifestyle manifested by Pencillium marneffei. In: Heitman, J., Kronstad, J.W., Taylor, J.W., Casselton, our knowledge of the relationships between the species. Further, L.A. (Eds.), Sex in Fungi: Molecular Determination and Evolutionary it provides important new information on the reproductive mode Implications. ASM Press, Washington, DC, pp. 201–212. of this group. Altogether our results suggest that sex has been lost Galagan, J.E., Calvo, S.E., Cuomo, C., Ma, L.J., Wortman, J.R., Batzoglou, S., Lee, S.I., Basturkmen, M., Spevak, C.C., Clutterbuck, J., Kapitonov, V., Jurka, J., Scazzocchio, recently in the Penicillium species, although further studies are still C., Farman, M., Butler, J., Purcell, S., Harris, S., Braus, G.H., Draht, O., Busch, S., needed to reject the hypothesis of cryptic sex. If sex has really been D’Enfert, C., Bouchier, C., Goldman, G.H., Bell-Pedersen, D., Griffiths-Jones, S., lost in the Penicillium species, our analyses suggest at least ten Doonan, J.H., Yu, J., Vienken, K., Pain, A., Freitag, M., Selker, E.U., Archer, D.B., Penalva, M.A., Oakley, B.R., Momany, M., Tanaka, T., Kumagai, T., Asai, K., independent origins of asexuality, which makes this group a model Machida, M., Nierman, W.C., Denning, D.W., Caddick, M., Hynes, M., Paoletti, M., of choice to investigate the maintenance of sex. Fischer, R., Miller, B., Dyer, P., Sachs, M.S., Osmani, S.A., Birren, B.W., 2005. Sequencing of Aspergillus nidulans and comparative analysis with A. fumigatus and A. oryzae. Nature 438, 1105–1115. Acknowledgments Giraud, T., Yockteng, R., López-Villavicencio, M., Refregier, G., Hood, M.E., 2008. Mating system of the anther smut fungus Microbotryum violaceum: selfing We thank two anonymous reviewers and Matthew Fisher for under heterothallism. Eukaryot. Cell 7, 765–775. their comments on previous versions of this manuscript, and John Glass, N.L., Donaldson, G.C., 1995. Development of primer sets designed for use with the PCR to amplify conserved genes from filamentous ascomycetes. Appl. Taylor for stimulating discussions. We thank Cony Decock and Lu- Environ. Microbiol. 61, 1323–1330. cile Sage, curators of the MUCL and CMPG collections respectively, Glass, N.L., Jacobson, D.J., Shiu, P.K.T., 2000. The genetics of hyphal fusion and for providing numerous isolates. This work was supported by the vegetative incompatibility in filamentous ascomycete fungi. Ann. Rev. Genet. 34, 165–186. ‘‘Consortium National de Recherche en Génomique”, and the ‘‘Ser- Guindon, S., Gascuel, O., 2003. A simple, fast, and accurate algorithm to estimate vice de Systématique Moléculaire” of the Muséum National d’His- large phylogenies by maximum likelihood. Syst. Biol. 52, 696–704. toire Naturelle (IFR 101). It is part of the agreement no. 2005/67 Haedens, V., Malagnac, F., Silar, P., 2005. Genetic control of an epigenetic cell degeneration syndrome in Podospora anserina. Fungal Genet. Biol. 42, 564–577. between the Genoscope and the Muséum National d’Histoire Hoff, B., Pöggeler, S., Köck, U., 2008. Eighty years after its discovery, Fleming’s Naturelle on the Project ‘‘Macrophylogeny of life” directed by Guil- Penicillium strain discloses the secret of its sex. Eukaryot. Cell 7, 465–470. laume Lecointre. We acknowledge the Grant ANR 09-BLAN-Fungi- Horn, B.W., Moore, G.G., Carbone, I., 2009a. Sexual reproduction in Aspergillus flavus. Mycologia 101, 423. Sex. DMdV benefited from the support of the Chair Modélisation Horn, B.W., Ramirez-Prado, J.H., Carbone, I., 2009b. Sexual reproduction and Mathématique et biodiversité VEOLIA-Ecole Polytechnique-MNHN. recombination in the aflatoxin-producing fungus Aspergillus parasiticus. Fungal Genet. Biol. 46, 169–175. Hull, C.M., Raisner, R.M., Johnson, A.D., 2000. Evidence for mating of the ‘‘asexual” Appendix A. Supplementary material yeast Candida albicans in a mammalian host. Science 289, 307–310. James, T.Y., Kauff, F., Schoch, C.L., Matheny, P.B., Hofstetter, V., Cox, C.J., Celio, G., Supplementary data associated with this article can be found, in Gueidan, C., Fraker, E., Miadlikowska, J., Lumbsch, H.T., Rauhut, A., Reeb, V., Arnold, A.E., Amtoft, A., Stajich, J.E., Hosaka, K., Sung, G.H., Johnson, D., O’Rourke, the online version, at doi:10.1016/j.fgb.2010.05.002. B., Crockett, M., Binder, M., Curtis, J.M., Slot, J.C., Wang, Z., Wilson, A.W., Schussler, A., Longcore, J.E., O’Donnell, K., Mozley-Standridge, S., Porter, D., References Letcher, P.M., Powell, M.J., Taylor, J.W., White, M.M., Griffith, G.W., Davies, D.R., Humber, R.A., Morton, J.B., Sugiyama, J., Rossman, A.Y., Rogers, J.D., Pfister, D.H., Hewitt, D., Hansen, K., Hambleton, S., Shoemaker, R.A., Kohlmeyer, J., Aanen, D.K., Hoekstra, R.F., 2007. Why sex is good: on fungi and beyond. In: Volkmann-Kohlmeyer, B., Spotts, R.A., Serdani, M., Crous, P.W., Hughes, K.W., Heitman, J., Kronstad, J.W., Taylor, J.W., Casselton, L.A. (Eds.), Sex in Fungi: Matsuura, K., Langer, E., Langer, G., Untereiner, W.A., Lucking, R., Budel, B., Molecular Determination and Evolutionary Implications. ASM Press, Geiser, D.M., Aptroot, A., Diederich, P., Schmitt, I., Schultz, M., Yahr, R., Hibbett, Washington, DC, pp. 527–534. D.S., Lutzoni, F., McLaughlin, D.J., Spatafora, J.W., Vilgalys, R., 2006. Aguileta, G., Marthey, S., Chiapello, H., Lebrun, M.H., Rodolphe, F., Fournier, E., Reconstructing the early evolution of Fungi using a six-gene phylogeny. Gendrault-Jacquemard, A., Giraud, T., 2008. Assessing the performance of Nature 443, 818–822. single-copy genes for recovering robust phylogenies. Syst. Biol. 57, 613–627. Judson, O.P., Normark, B.B., 1996. Ancient asexual scandals. Trends Ecol. Evol. 11, Berbee, M.L., Yoshimura, A., Sugiyama, J., Taylor, J.W., 1995. Is Penicillium 41–46. monophyletic? An evaluation of phylogeny in the family Trichocomaceae Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of from 18S, 5. 8S and ITS ribosomal DNA sequence data. Mycologia 87, 210–222. the evolutionary tree topologies from DNA sequence data, and the branching Billiard, S., López-Villavicencio, M., Devier, B., Hood, M.E., Fairhead, C., Giraud, T., in order in hominoidea. J. Mol. Evol. 29, 170–179. press. Having sex, yes, but with whom? Inferences from fungi on the evolution Kuhner, M.K., Felsenstein, J., 1994. A simulation comparison of phylogeny of anisogamy and mating types. Biol. Rev. algorithms under equal and unequal evolutionary rates. Mol. Biol. Evol. 11, Birky, C.W., 1999. An even broader perspective on sex and recombination. J. Evol. 459–468. Biol. 12, 1013–1016. Le Gac, M., Hood, M.E., Fournier, E., Giraud, T., 2007. Phylogenetic evidence of host- Bruggeman, J., Debets, A.J.M., Wijngaarden, P.J., deVisser, J.A.G.M., Hoekstra, R.F., specific cryptic species in the anther smut fungus. Evolution 61, 15–26. 2003. Sex slows down the accumulation of deleterious mutations in the Lin, X., Heitman, J., 2007. Mechanisms of homothallism in Fungi and transitions homothallic fungus Aspergillus nidulans. Genetics 164, 479–485. between heterothallism and homothallism. In: Heitman, J., Kronstad, J.W., Burt, A., Carter, D.A., Koenig, G.L., White, T.J., Taylor, J.W., 1996. Molecular markers Taylor, J.W., Casselton, L.A. (Eds.), Sex in Fungi: Molecular Determination and reveal cryptic sex in the human pathogen Coccidioides immitis. Proc. Natl. Acad. Evolutionary Implications. ASM Press, Washington, DC, pp. 35–57. Sci. USA 93, 770–773. 706 M. López-Villavicencio et al. / Fungal Genetics and Biology 47 (2010) 693–706

Lobuglio, K.F., Pitt, J.I., Taylor, J.W., 1993. Phylogenetic analysis of two ribosomal Takada, M., Udagawa, S., 1988. A new species of heterothallic Talaromyces. DNA regions indicates multiple independent losses of a sexual Talaromyces state Mycotaxon 31, 417–425. among asexual Penicillium species in subgenus Biverticillium. Mycologia 85, Taylor, J.W., Jacobson, D.J., Fisher, M.C., 1999. The evolution of asexual fungi: 592–604. reproduction, speciation and classification. Annu. Rev. Phytopat. 37, 197– Marthey, S., Aguileta, G., Rodolphe, F., Gendrault, A., Giraud, T., Fournier, E., Lopez- 246. Villavicencio, M., Gautier, A., Lebrun, M.H., Chiapello, H., 2008. FUNYBASE: a Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. Clustal-W – improving the FUNgal phYlogenomic dataBASE. BMC Bioinf. 9, 456. sensitivity of progressive multiple sequence alignment through sequence Notredame, C., Higgins, D.G., Heringa, J., 2000. T-coffee: a novel method for fast and weighting, position-specific gap penalties and weight matrix choice. Nucleic accurate multiple sequence alignment. J. Mol. Biol. 302, 205–217. Acids Res. 22, 4673–4680. Nye, T.M., Lio, P., Gilks, W.R., 2006. A novel algorithm and web-based tool for Van Reenen-Hoekstra, E.S., Frisvad, J.C., Samson, R.A., Stolk, A.C., 1990. The comparing two alternative phylogenetic trees. Bioinformatics 22, 117–119. Penicillium funiculosum complex-well defined species and problematic taxa. O’Donnell, K., Ward, T.J., Geiser, D.M., Kistler, H.C., Aoki, T., 2004. Genealogical In: Samson, R.A., Pitt, J.I. (Eds.), Modern Concepts in Penicillium and Aspergillus concordance between the mating type locus and seven other nuclear genes Classification. Plenum Press, New York, pp. 173–192. supports formal recognition of nine phylogenetically distinct species within the Verna, J., Ballester, R., 1999. A novel role for the mating type (MAT) locus in the Fusarium graminearum clade. Fungal Genet. Biol. 41, 600–623. maintenance of cell wall integrity in Saccharomyces cerevisiae. Mol. Gen. Genet. O’Gorman, C.M., Fuller, H.T., Dyer, P.S., 2009. Discovery of a sexual cycle in the 261, 681–689. opportunistic fungal pathogen Aspergillus fumigatus. Nature 457, 471–474. White, T.J., Bruns, T., Lee, S., Taylor, J.W., 1990. Amplification and direct sequencing Otto, S.P., Lenormand, T., 2002. Resolving the paradox of sex and recombination. of fungal ribosomal RNA genes for phylogenetics. In: Innis, M.A., Gelfand, D.H., Nat. Rev. Genet. 3, 252–261. Sninsky, J.J., White, T.J. (Eds.), PCR Protocols: A Guide to Methods and Pagel, M., Meade, A., Barker, D., 2004. Bayesian estimation of ancestral character Applications. Academic Press Inc., New York, pp. 315–322. states on phylogenies. Syst. Biol. 53, 673–684. Wik, L., Karlsson, M., Johannesson, H., 2008. The evolutionary trajectory of the Peterson, S.W., 2000. Phylogenetic analysis of Penicillium species based on ITS and mating-type (mat) genes in Neurospora relates to reproductive behavior of taxa. LSU-rDNA nucleotide sequences. In: Samson, R.A., Pitt, J.I. (Eds.), Integration of BMC Evol. Biol. 8, 109. Modern Taxonomic Methods for Penicillium and Aspergillus Classification. Woo, P.C., Chong, K.T., Tse, H., Cai, J.J., Lau, C.C., Zhou, A.C., Lau, S.K., Yuen, K.Y., 2006. Hargrove Academic Publishers, Amsterdam. Genomic and experimental evidence for a potential sexual cycle in the Peterson, S.W., Bayer, E.M., Wicklow, D.T., 2004. Penicillium thiersii, Penicillium pathogenic thermal dimorphic fungus Penicillium marneffei. FEBS Lett. 580, angulare and Penicillium decaturense, new species isolated from wood-decay 3409–3416. fungi in North America and their phylogenetic placement from multilocus DNA Wu, J., Saupe, S.J., Glass, N.L., 1998. Evidence for balancing selection operating at the sequence analysis. Mycologia 96, 1280–1293. het-c heterokaryon incompatibility locus in a group of filamentous fungi. Proc. Pitt, J.I., 1979. Genus Penicillium and its Telemorphic States Eupenicillium and Natl. Acad. Sci. USA 95, 12398–12403. Talaromyces. Academic Press, London. Yang, Z., 1997. PAML: a program package for phylogenetic analysis by maximum Pöggeler, S., 2001. Mating-type genes for classical strain improvements of likelihood. Comput. Appl. Biosci. 13, 555–556. ascomycetes. Appl. Microbiol. Biotechnol. 56, 589–601. Yang, Z., 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol. Pöggeler, S., Hoff, B., Kuck, U., 2008. Asexual cephalosporin C producer Acremonium Evol. 24, 1586–1591. chrysogenum carries a functional mating type locus. Appl. Microbiol. Biotechnol. Yang, Z.H., 1998. On the best evolutionary rate for phylogenetic analysis. Syst. Biol. 74, 6006–6016. 47, 125–133. Posada, D., Crandall, K.A., 2001. Evaluation of methods for detecting recombination Yang, Z.H., Nielsen, R., 1998. Synonymous and nonsynonymous rate variation in from DNA sequences: computer simulations. Proc. Natl. Acad. Sci. USA 98, nuclear genes of mammals. J. Mol. Evol. 46, 409–418. 13757–13762. Yang, Z.H., Nielsen, R., 2002. Codon-substitution models for detecting molecular Raper, K.B., Thom, C., 1949. A Manual of the Penicillia. Williams and Wilkins, adaptation at individual sites along specific lineages. Mol. Biol. Evol. 19, 908– Baltimore, Maryland. 917. Robinson, D.F., Foulds, L.R., 1981. Comparison of phylogenetic trees. Math. Biosci. Yokoyama, E., Yamagishi, K., Hara, A., 2003. Structures of the mating-type loci of 53, 131–147. Cordyceps takaomontana. Appl. Environ. Microbiol. 69, 5019. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with Yun, S.H., Berbee, M.L., Yoder, O.C., Turgeon, B.G., 1999. Evolution of the fungal self- applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. fertile reproductive life style from self-sterile ancestors. Proc. Natl. Acad. Sci. Shimodaira, H., Hasegawa, M., 2001. CONSEL: for assessing the confidence of USA 96, 5592–5597. phylogenetic tree selection. Bioinformatics 17, 1246–1247. Zhang, J.Z., Nielsen, R., Yang, Z.H., 2005. Evaluation of an improved branch-site Steinberg-Neifach, O., Eshel, D., 2000. Simultaneous expression of both MAT loci in likelihood method for detecting positive selection at the molecular level. Mol. haploid cells suppresses mutations in yeast microtubule motor genes. Mol. Gen. Biol. Evol. 22, 2472–2479. Genet. 264, 300–305.